AMD opens Opteron to all some ~ and IBM signs up

AMD has expanded its friends and family program in an obvious way by allowing partners to plug their CPUs, accelerators and other components directly into Opteron processor sockets. AMD's unveiling today of Torrenza 2 - aka the Torrenza Innovation Socket - builds on an existing deal that let third parties tap into the company's Hypertransport technology. Now, AMD has agreed to open up the entire Opteron socket specifications to preferred partners willing to pay a licensing fee. As a result, customers could start seeing some odd but fantastic motherboards with, for example, a mix of Opteron and Cell chips or even an UltraSPARC T1/Opteron combo.

The Torrenza brand may sound silly, but the concept behind it is sound. AMD has proved more willing than Intel to reveal its hardware IP to partners, and, in return, partners have shown great interest in flaunting their wares via Hypertransport. Thus far, Intel has been reluctant to free up its Northbridge in a similar manner or to promise a more open architecture when it moves to Opteron copy chips outfitted with its CSI technology. The market for add-on components for Opteron chips is young but promising. Companies such as Pathscale lauched their InfiniPath HTX Adapter for AMD's architecture first and then followed with Intel gear that plugs into PCI Express. DRC and XtremeData have taken the process one step further with FPGAs that already plug directly into Opteron sockets. The companies, however, had to reverse engineer AMD's design.

Now companies such as Sun, IBM, Cray and Fujitsu have agreed to pay for an Opteron socket license.

Not everyone can access AMD's program, making it open but not wholly open.

"We will want to open it up and license it to people that have the capability and staying power and support that meets the enterprise standards associated with Opteron," AMD SVP Marty Seyer told us.

IBM and AMD have already teamed on a supercomputing win at Los Alamos Lab where they'll be combining Opteron- and Cell-based servers. In the future, such chips could sit together on the same motherboard. Likewise, Sun could put its mutli-core UltraSPARC T1 chips on an Opteron motherboard, giving the industry standard server market some serious flash.

DRC's CEO Larry Laurich applauded AMD's move.

"This invites more competition in one sense, but I think it benefits the market as a whole," Laurich said. "I think AMD is being smart and aggressive."

XtremeData, which also makes FPGA accelerators for Opteron sockets, already has plans in place to deliver a new product for AMD's Socket F designs. It should start shipping gear in the next six months, said CEO Ravi Chandran.

"Everything is going to Socket F very soon," he said. "That has always been on our roadmap."

The company plans to release some novel server appliances for the data warehousing and medical imaging markets. The systems will have Opteron chips and FPGAs on the same motherboards, along with the Postgres database.

The presence of all these accelerators and add-on components is adding some real oomph to the x86 server market. It will be interesting to see how Intel responds. ®

And today:

IBM's Power7 chip going into Opteron motherboards

***************************************

By Ashlee Vance in Mountain View (ashlee.vance@theregister.co.uk)

Published Friday 22nd September 2006 17:28 GMT

Exclusive AMD has secured yet another major partner win thanks to Opteron. Starting with the Power7 processor, IBM will give up on making its own Unix/RISC box motherboards. Instead, it will plug the Power chips directly into slightly modified Opteron boards in an effort to save money.

Neither IBM nor AMD would comment for this story. Multiple sources, however, did confirm the move in interviews with The Register. IBM and AMD have already signed an agreement around the arrangement.

Our sources have also revealed that Sun Microsystems is in discussions with AMD to pursue a similar plan for its UltraSPARC and UltraSPARC T1 processors.

"We are excited about AMD's common socket initiative because it opens up a whole new set of possibilities in systems design, but we aren't prepared to discuss any specific products using this at this time," said Sun's server chief John Fowler.

This week, AMD revealed a new partner program that will let third-parties such as Sun, IBM and smaller vendors build products that fit directly into Opteron sockets. Most analysts believed that IBM and Sun were just experimenting with how they would make use of this Opteron option. But The Register can confirm that engineering efforts have started within IBM to fit the far off Power7 chips right into the Opteron sockets.

By playing off Opteron motherboards, IBM would enjoy some serious cost savings. It would no longer need to produce separate motherboards for its Unix server line. Some questions remain as to how well x86 motherboards will stack up against RISC boards in the high-end SMP server market. Although, experts interviewed for this story said that by the time Power7 arrives - possibly in 2009 - AMD should be able to churn out top-notch SMP systems.

"I think people are getting smarter about how to take advantage of commodity stuff," said Fred Weber, AMD's former CTO and now CEO of MetaRam.

Given the apparent direction of Opteron designs, it would be feasible to create 32-socket systems outfitted with four-core chips, Weber said. And that's just in the 2007 or 2008 timeframe. So, IBM could easily build large SMPs by the time Power7 ships. When asked if all the Tier1 RISC vendors are likely to head toward AMD's boards, Weber said, "It would not surprise me if they did or if they didn't."

There are clear cost benefits to be had here for the RISC crowd. Intel, for example, plans to have both Itanium and Xeon chips slot into the same motherboards by at least 2009 and possibly 2008. That would lower the overall costs of producing Itanium servers for OEMs and likely translate to a cost break for end customers. If AMD could sign up IBM, Sun and Fujitsu, then it would enjoy a significant volume play and associated cost savings. And all three of the RISC vendors would potentially save money by turning to a single supplier for their motherboards rather than producing custom gear on their own.

Beyond the cost aspects, AMD has to be pleased from a public relations standpoint to secure this deal with IBM and to have possible deals with Sun and Fujitsu in the works. Its pay for openness policy seems to be working well against Intel for the moment.

IBM has spent the last few years pouring money into specialized chipsets for Xeon-based servers. Big Blue now seems to be forming deeper ties with AMD. ®

Did some looking around....found this:

IBM's Power5 processor worth a second look

**********************AN EXCERPT *******************************

If all things were equal and IBM made its systems as accessible as Dell and Hewlett-Packard do theirs, the IBM Power5 processor could bury Intel’s Itanium 2. First introduced last summer, the Power5 is a one-two punch, a triumph of engineering from a company that excels not only in processor design but also in the submicron science of chip manufacturing and packaging.

The Power5 is plenty fast, of course. But it can also be viewed as IBM’s first serious attempt to meet customers’ needs beyond speed. The Power5 offers improved power efficiency and terrific scalability, supports non-IBM operating systems (including Linux and Windows), and delivers partitioning and virtualization unmatched by current Intel technology.

The Power5 also foreshadows a new generation of 64-bit, PowerPC-based workstations and servers from IBM’s longtime partner in Power, Apple Computer. And IBM recently pulled an unexpected move for a company built on patents by publishing the Power architecture and tools under an open license.

There are so many ways in which Power5’s influence reaches beyond IBM’s primary base of well-heeled customers. Although IBM also sells Itanium 2, Opteron, and Xeon servers, the company seems clearly intent on putting Power5 systems in the hands of Linux and Windows administrators. Whether that makes sense will be up to customers, but the sheer technical muscle of Power5 and the faltering fortunes of the Itanium architecture demand IBM’s flagship processor take a trip under our microscope.

Power secrets

IBM has consistently attracted the brightest minds, the kind of engineers who deserve the moniker “computer scientist.” In the 1980s, these scientists cooked up a processor architecture that was built for performance: the IBM 801, the original RISC processor. The 801’s legacy lives on in the IBM Power series of enterprise-class processors.

The major difference between a RISC processor and a CISC processor, such as Intel’s x86, can be viewed as a tug-of-war between programmers and chip designers. CISC processors are designed to make application developers’ lives easier by reducing common operations to single, long-executing native instructions, giving CISC a reputation as a slow but friendly design. Compared in that light, RISC is fast and unfriendly. Each of its simple instructions serves a very narrow purpose, executes quickly, and parallelizes exceptionally well. RISC requires patient, gifted programmers and meticulously optimized compilers; RISC’s success attests to an abundance of both.

The best known Power5 attribute is its integration of two discrete RISC cores on a single chip. Announcements from AMD, Intel, and Sun Microsystems regarding upcoming multicore processors focused attention on this aspect of Power5, but multicore was also a feature of its predecessors, Power4 and Power4+. According to IBM, Power5 is fully compatible with Power4 executables. The wonder of multicore is that it delivers the pipe dream of more speed in less space without a marked increase in heat. But as you’ll see, multicore is not simply SMP on a chip.

Click for larger view.

For one thing, the Power5’s cores share a very fast Level 2 cache. The speed and quantity of cache is a factor in the performance of all microprocessors. (The evolution of the x86 shows Intel to be utterly cache-obsessed.) With simple instructions flying through a RISC CPU so rapidly, the cache’s efficiency in reducing the number of trips to RAM becomes the key to the whole design.

Continued
1 | 2 | 3 | Next Page »

Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.